1. Field of the Invention
The present invention relates to an image sensor and a method of driving a transfer transistor of the image sensor that transfers photon-induced charges, and more particularly, to an image sensor and a method of driving a transfer transistor of the image sensor that are capable of maintaining a depletion degree of charges in a photodiode when the photodiode is reset.
2. Discussion of Related Art
Image sensors may be classified into a charged coupled device (CCD) sensor and a complementary metal oxide semiconductor (CMOS) image sensor, which use electron-hole pairs separated by light having an energy higher than a silicon band gap, in which an amount of irradiated light is generally estimated by accumulating either electrons or holes.
The CMOS image sensor includes image pixels each having a photodiode and transistors, similar to a typical CMOS device. Image-signal processing and detecting circuits can be integrated in an external block of the pixel. This eliminates a need for an image-signal processing circuit included in a separate chip, allows a variety of image sensor structures to be adopted, and provides flexibility so that subsequent image processing is performed by hardware.
A 4-transistor pixel structure widely used to implement a CMOS image sensor is shown in FIG. 1. The 4-transistor pixel structure is composed of four transistors. A photodiode PD that is a photo-sensing unit and four NMOS transistors constitute one unit pixel. Among the four NMOS transistors, a transfer transistor Tx serves to transfer photo charges generated by the photodiode PD to a diffusion node region 131, a reset transistor Rx serves to discharge charges from the diffusion node region 131 or the photodiode PD so that a signal is detected, a drive transistor Dx serves as a source follower transistor, and a switching transistor Sx is used for switching/addressing. The transfer transistor Tx may be implemented by a gate, a gate oxide layer, and a p-type substrate, the photodiode PD may be generally implemented by an n- or no-doped region and a surface p-doped region, and the diffusion node 131 may be implemented by an n+ doped region.
In FIG. 1, the photodiode PD receiving light and a capacitor 118 connected parallel to the photodiode PD constitute a light-receiving unit, and the transfer transistor Tx serves to transfer electrons generated by photons to the diffusion node 131.
The transfer transistor Tx serves as a transmission channel that transfers electrons generated from a surface of the photodiode PD to the diffusion node 131 or performs a reset function to completely remove electrons from the photodiode PD in response to a voltage applied to its gate. The diffusion node 131 includes a diffusion capacitor 114 and a gate capacitor of the drive transistor Dx. The diffusion node 131 is reset by the reset transistor Rx. That is, the diffusion node 131 is reset before receiving the electrons from the photodiode PD region, or a reset voltage is applied to the diffusion node 131 to reset the photodiode PD region. A voltage is applied to a gate 141 of the switching transistor Sx in order to select one row for a two-dimensional image. Each pixel is biased by a current source 150, which activates the drive transistor Dx and the switching transistor Sx so that a voltage at the diffusion node 131 is read out to an output node 142.
In a CMOS image sensor having the 4-transistor pixel shown in FIG. 1, photon-induced carriers accumulated in the photodiode after the photodiode is reset are transferred to the floating diffusion node, causing a voltage drop across the diffusion node, and thus the voltage drop is used to detect the amount of the photon-induced carriers. In this case, the transfer transistor must perform uniform reset and transfer operations in order to accurately and uniformly detect the amount of the accumulated photon-induced carriers. A variety of structures of a conventional 4-transistor pixel including a fully reset pinned photodiode to allow a transfer transistor to perform uniform reset and transfer operations are disclosed. The pinned photodiode uses a state where all movable charges in the photodiode are completely depleted and a voltage is not changed any more. Ideally, a photodiode voltage is always pinned into a constant value irrespective of an external bias such as a voltage at the floating diffusion node. Therefore, the reset and transfer conditions for the transfer transistor becomes always constant.
However, in the conventional CMOS image sensor having a 4-transistor pixel, a reduced operating voltage or a changed process condition may always change reset and transfer conditions depending on a relationship between the gate voltage of the transfer transistor and the voltage at the floating diffusion node.
Specifically, in a conventional driving method using a power supply voltage (VDD) as a transistor turn-on voltage, when the transfer transistor is reset, a voltage at the floating diffusion node is equal to the gate voltage VDD of the reset transistor minus a threshold voltage value threshold voltage (Vth) of the reset transistor RX (VDD-Vth). This value automatically allows a difference between the gate voltage VDD of the transfer transistor and the voltage at the floating diffusion node to be equal to the threshold voltage Vth. Generally, since the reset and transfer transistors are formed in the same doping condition on a substrate, they have a similar threshold voltage Vth. In this case, a state in the condition corresponds to an edge between a pinch-off state region, in which a transfer transistor's edge at the floating diffusion node begins to be turned on according to the definition of the threshold voltage Vth, and a linear operation region. At a time when the transfer transistor's edge at the floating diffusion node is turned on, a certain amount of electrons may promptly move from the floating diffusion node to the channel region of the transfer transistor. Accordingly, the voltage at the floating node is significantly changed due to the capacitance. Furthermore, the amount of electrons from the floating diffusion node significantly changes in a small difference in threshold voltage between the transfer transistor and the reset transistor. Such a nonuniform amount of electrons from the floating diffusion node causes irregularity of the reset condition, thus deteriorating the quality of an image.
Unstable reset and transfer operations of the transfer transistor may cause two typical problems of increased dark current and increased fixed pattern noise.
In the reset operation, since the reset transistor Rx is turned on the floating diffusion node has a low impedance with respect to the ground, the voltage is substantially the same VDD-Vth as the power supply voltage VDD. In the transfer operation, since the reset transistor Rx is turned off and the floating diffusion node has a high impedance with respect to the ground, electrons in the channel of the transfer transistor flow into the floating node (clock feedback), so that the voltage at the floating node becomes lower than the voltage VDD-Vth. Additionally, the gate voltage of the transistor increases an ON voltage according to boosting condition. In this process, the floating node voltage differs between the reset and transfer operations. This different voltage conditions have not caused any trouble because a completely depleted (i.e., completely reset) pinned photodiode is employed, i.e., the pixel is driven after the photodiode is completely depleted. The use of the pinned photodiode can also suppress dark current and other noises.
However, as a modern semiconductor process and device is scaled down and an operating voltage is reduced, the floating diffusion node voltage gets gradually lower. Accordingly, a pinning voltage of a pinned photodiode gets lower, thereby deteriorating a pixel characteristic such as well capacity.
Further, a voltage barrier necessarily exists between the pinned photodiode and the channel of the transfer transistor to some extent. To suppress the effect of the barrier when the transfer transistor is turned on, a pinning voltage is made significantly different from the voltage at the floating diffusion node. When the barrier is not sufficiently reduced, the pinned photodiode is not completely reset, which may cause more severe problems. That is, when an operating voltage indicated as the power supply voltage VDD is reduced, a difference between the pinning voltage and the floating diffusion node voltage is reduced. In addition, the well capacity may be lowered and resetting (e.g., depletion) may be insufficient.
To solve the problems, in a conventional technique, a voltage at a floating diffusion node forcibly rises from a typical voltage VDD-VTH to the power supply voltage VDD using a boosting circuit. In another conventional technique, the floating diffusion node voltage rises to the power supply voltage VDD sufficiently and quickly using a reset transistor Rx of a PMOS type, not a conventional NMOS type.
However, the voltage boosting circuit applies a voltage over a normal operation condition, which may degrade the reliability of a gate oxide. When a PMOS transistor is used as a reset transistor Rx, it occupies a wider area than an NMOS transistor. Accordingly, a fill factor is reduced to deteriorate a characteristic of the device, and two times more noise than in an NMOS transistor is generated, as known in the art. Further, this approach has a limitation of characteristic enhancement in a complete reset condition.